SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Handshake AI is hiring an AI Red Teamer specializing in CBRNE (chemical, biological, radiological, nuclear, and explosive) threat evaluation. This role sits at the intersection of AI safety and national security, working directly with frontier AI labs to stress-test model defenses against sophisticated adversarial scenarios.
You will design technically grounded adversarial prompts to evaluate whether AI models appropriately refuse, hedge, or redirect queries related to CBRNE threats. Your work involves probing for dangerous knowledge gaps in safety guardrails—testing whether models can be manipulated into providing meaningful uplift toward creation, acquisition, or deployment of weapons or hazardous materials. This is not about generating harmful content, but identifying and documenting model failures so labs can fix them before deployment.
Day-to-day responsibilities include: designing adversarial prompts that test model responses; evaluating technical accuracy and real-world consequence of outputs; probing dual-use knowledge boundaries; testing multi-step attack chains; scoring responses against harm taxonomies; documenting findings with clear technical reasoning; distinguishing between publicly available information and operationally significant uplift; contributing to CBRNE-specific evaluation frameworks; collaborating with red teamers, AI researchers, and policy teams; and staying current on model capabilities and jailbreak techniques.
Required qualifications: graduate-level education or equivalent professional experience in a CBRNE field (chemistry, biochemistry, microbiology, virology, nuclear physics, radiochemistry, materials science, munitions/ordnance, chemical engineering, or related); ability to evaluate technical accuracy and real-world consequences; understanding of dual-use research concerns; hands-on experience with multiple LLMs; creative adversarial problem-solving; clear written communication; strong ethical judgment; self-directed and collaborative approach.
Nice-to-have qualifications include active/prior security clearance, threat assessment or WMD analysis experience, biosafety/biosecurity background, familiarity with regulatory frameworks (CWC, BWC, IAEA, ATF), red teaming or penetration testing experience, Python/scripting skills, published research, or prior trust and safety/AI evaluation work.
Handshake AI has grown from $0 to ~$1B run rate in 2025, working with frontier AI labs, Fortune 500 partners, and educational institutions. The team includes engineers and scientists from Palantir, Meta, Scale AI, and former YC founders.
